On Friday 15 June 2007 09:57, Michel Meyers wrote: > Kern Sibbald wrote: > > Hello, > > > > If you are a user of Bacula on Windows, I would be interested in your > > responses to the following: > > > > 1. I am considering to change the default installation location for Bacula on > > Windows to be the same as it was previously -- that is the \bacula directory > > on the main disk. The current installation places files in the "standard" > > Windows locations, but IMO, it is very inconvenient because they are > > extremely long names with spaces that are very hard to remember, and are > > sprayed all over the disk. > > I definitely prefer the 'old' way of putting everything in c:\bacula and > have kept my FD installations that way when the new installer came out. > I don't care what M$ says, having the program in a directory that is not > OS language specific, doesn't contain spaces and is self contained > offers me certain advantages over their Program Files + Documents and > Settings + Registry mess, namely: > > - Language independant paths: I can use external scripts to update > Bacula. No need to push an installer and somehow coerce the Windows box > to run it so that it can resolve the pathnames 'locally': I can just > give an RPC command to shut down the FD, map the drive, place the new > files in the standard directory and have the FD service come up through > RPC again.) > - No spaces in filenames: Obviously this avoids many issues with having > to quote pathnames in config files etc etc. > - Everything in/under one directory: After running the uninstaller, > ensuring that everything is gone is easily done by wiping said > directory. Similarly, finding the configuration files for the program is > much easier that way than having to wade through Documents and > Settings\<variable_username>\(Local Settings\)Application Data.
Yes, in the beginning when this new Windows standard layout was proposed, I thought it was a reasonable idea, but for the reasons that you mentioned above, it is very complicated. Even worse, for languages where there is an accented character in one of these system names, the user must manually edit the conf files to make Bacula work because (apparently) the environment variable that defines the directory path is encoded in UTF-16, while Bacula requires UTF-8 in the conf files :-( > > Apart from "it's what M$ wants", I do not see any advantage to actually > using their convention (unless people use roaming profiles, but I doubt > Bacula is a candidate for that). At least the config files of a program > should always stay with it. That's my, very biased, 2 Eurocents. ;) > > That said, as I'm only using the Win32 FD, the defaults in the installer > doesn't really bother me. I simply unpack the files from it using 7-Zip > and then build my own or manually install the FD. > > I don't think I can help with any of the other points, maybe the > regression scripts if that is doable in a VM and doesn't require any > hardware? Yes, I believe that it will run in a VM. It should be interesting to try anyway, and the scripts can be run in disk only mode without the need for a tape drive. Even though it is better to run the tape tests, it would be a big step forward if even the disk based regression tests were run. If you have the time, I can see several projects: 1. Examine the install script, and modify it to allow the user to choose between a single directory and the "standard" Windows setup that is currently used. This shouldn't be *too* difficult, because one can pull the old code from the SVN that wrote it all to one directory. 2. Possibly enhance the install script to allow the user to select advanced install or something and then to specify the path to all the various parts that are currently configurable (i.e. scripts, binaries, working directory, ...) 3. Try running the regression scripts. Ideally several people would do this together. In the long run, I would like the regression scripts to be run at a number of different places. For example, one regression tester could concentrate on SQLite, and PostgreSQL, another tape tests, AMD, i386, WinXP, Win2003 server, Vista ... This is not a area where running regression tests is lost effort due to "duplication" of effort. In any case, any help you can give, and any input on the process would be appreciated as I would like to see the Win32 testing effort brought up to the same levels as the Linux/Solaris/FreeBSD. Even though, I personally prefer Linux environments, Win32 is important, and I would like to see it properly supported by the Bacula project. Regards, Kern > > Greetings, > Michel > ------------------------------------------------------------------------- This SF.net email is sponsored by DB2 Express Download DB2 Express C - the FREE version of DB2 express and take control of your XML. No limits. Just data. Click to get it now. http://sourceforge.net/powerbar/db2/ _______________________________________________ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
